Link-local addresses and advertised prefixes
A node can first form a link-local address, while a host can combine suitable advertised prefixes with interface identifiers for other addresses. Lengths must satisfy the applicable address and link rules, and duplicate-address checks normally precede use.
Stateless contrasts with maintaining a server allocation binding for every address. Hosts still retain addresses and lifetimes, and routers advertise configuration. Generating an address alone does not establish every route or replace default-router discovery.
Preferred and valid lifetimes differ
An address becomes deprecated when its preferred lifetime expires: it is generally avoided for new communication, while existing communication can continue within its valid lifetime. Expiration of valid lifetime is a stronger boundary, not the same timer.
Keeping old and new addresses on an interface supports gradual renumbering. Advertisement-driven lifetime changes have validation and protection rules; an arbitrary short announcement does not unconditionally delete every existing address immediately.
References: [1]
Identifiers and configuration choices can evolve
SLAAC does not require a permanent hardware address to be embedded in IPv6 identifiers. RFC 8064 recommends opaque stable generation, while RFC 8981 specifies rotating temporary addresses with a different stability goal. Neither alone guarantees anonymity of all activity.
SLAAC can coexist with DHCPv6. RFC 9762 adds a P flag expressing a preference for prefix delegation to clients supporting that capability; other clients ignore it. An early Autonomous flag rule alone is therefore not a universal description of every modern configuration path.